Re: [Hecl-devel] Updated repository for Hecl

[David N. W. <da...@de...>]

Wolfgang S. Kechel wrote:

> The build process has gundergone a major overhaul and should be tested
> in various environments for various targets. The build result is now
> placed in platform-specific subdirectories underneath .../hecl/build.

Hrm... this is missing:

  [taskdef] Could not load definitions from resource antenna.properties.
It could not be found.

I have an idea: perhaps it's possible for ant to auto-generate a
properties file, because we need to accomplish two things:

1) People who download the sources should be able to compile things
without too much trouble.

2) We need to be able to compile our target projects without worrying
about stomping eachother's changes in version control - switching back
and forth between 1.0 and 1.1 as a target, for instance.  I'll
investigate how this could be done.   My idea: if it doesn't exist,
create it, so that it's not checked into subversion.

> I think we should move the prebuild jar files into a separate directory
> in the repository (maybe 'prebuild') and modify build.xml appropriatly.
> We also need to update INSTALL.

I will fix INSTALL when things have settled down just a bit.  How about
something like /bin or /lib for the checked in .jar's?  Some name that's
obvious would be best.  Maybe just jars/ ?

> Due to lack of time and knowledge, I was not able to check heclbuilder
> and someother targets, so I need to leave it for the community for now
> and postpone that for now.
> Updating javadoc is a task I did not want to touch for the moment. Maybe
> somebody else has some time left for this.

I will take care of those when the time comes (when we've got everything
else in place).

> There are some other upcoming changes, especially a more flexible http
> interface that allows to post query data, additional header fields (i.e
> cookies, session ids,...) and supports encoding-sets and charsets (for
> both mobile and desktop).

Cool!

> -------------Summary of changes -------------

> 	* build.xml: Modified build process, added more properties,
> 	simplified jar file generation. j2me versions robust now, applet
> 	target should be checked by somebody else.
> 	Bumped midlet version number to 1.1.
> 
> 	* core/org/hecl/Interp.java: setResult(boolean) now uses
> 	well-known static IntThings to create the result.
> 
> 	* core/org/hecl/HashCmds.java (http): Added new commands
> 	hkeys, hclear, hremove. Simplified code.

Looks good, although hclear could arguably be implemented as

proc hclear {varname} {
    upvar {
        unset $varname;
        set varname [hash {}]
    }
}

But, with the new system, it's probably almost cheaper in terms of bytes
to implement it directly, especially since we don't have a built in
library mechanism yet.

> 	* core/org/hecl/Ensemble.java: Added newline at end of file to
> 	avoid complaints of scn/diff.

This one is something Wojciech added, but it isn't actually used
anywhere.  I left it in, because it's not a bad idea... what do you
think about ensembles?  Some of the math commands would be a good
candidate to group under another command name, like "math sin 3432423"
or expr or something along those lines.

> 	* core/org/hecl/HashThing.java: Experimental new representation as
> 	string. Properly working converion to list, passes tests.
> 
> 	* core/org/hecl/ListThing.java: Experimental new implementation of get.

I'm not in favor of these without further discussion, as per the other
posting.

Also... with regards to MathCmds.java - is it possible to get Antenna to
only build the Java 1.5 commands if that's defined?  I'll see if I can
make it do that.

-- 
David N. Welton
- http://www.dedasys.com/davidw/

Linux, Open Source Consulting
- http://www.dedasys.com/